haz-query 0.1.0

//! Query-engine entry point and module map.
//!
//! `engine::execute` takes a loaded workspace, its validated task
//! graph, the bearing project (if any), and a typed
//! [`spec::QuerySpec`] covering every filter family (`QRY-003`
//! per-attribute, `QRY-004` relational, `QRY-005` boolean
//! shortcuts), and returns the matching tasks' `TaskId`s in
//! canonical `(ProjectName, TaskName)` order per `QRY-007`.
//!
//! `execute_non_relational` is a convenience wrapper that uses
//! an empty graph; it is appropriate only for specs that contain
//! no relational filters.

pub mod candidate;
pub mod filter;
pub mod relational;
pub mod spec;

use haz_dag::graph::TaskGraph;
use haz_domain::name::ProjectName;
use haz_domain::task_id::TaskId;
use haz_domain::workspace::Workspace;

use crate::engine::candidate::collect_candidates;
use crate::engine::filter::passes_non_relational;
use crate::engine::relational::{RelationalTargets, passes_relational};
use crate::engine::spec::{QueryError, QuerySpec};

/// Run the engine against a workspace and return the matching
/// task identities in canonical `(ProjectName, TaskName)` order.
///
/// Evaluates every filter family of `QRY-003`..`QRY-005` against
/// the candidate set determined by `bearing_project` per
/// `QRY`'s candidate-set vocabulary. Filters compose by
/// intersection per `QRY-006`.
///
/// `graph` MUST be the validated hard-edge graph (`DAG-008`) of
/// `workspace`; the engine consults it only to evaluate the four
/// `QRY-004` relational filters. Callers without any relational
/// filter in `spec` MAY pass `&TaskGraph::default()` or use
/// [`execute_non_relational`].
///
/// `bearing_project` is the cwd-derived bearing project name per
/// `EXEC-022`. It restricts the candidate set per `QRY`
/// vocabulary and supplies the canonicalisation root for any
/// project-relative `--inputs` / `--outputs` atoms.
///
/// # Errors
///
/// Returns [`QueryError`] when the bearing project does not
/// exist, when a project-relative atom is supplied without a
/// bearing project, or when the glob-glob intersection routine
/// fails.
pub fn execute(
    workspace: &Workspace,
    graph: &TaskGraph,
    bearing_project: Option<&ProjectName>,
    spec: &QuerySpec,
) -> Result<Vec<TaskId>, QueryError> {
    let candidates = collect_candidates(workspace, bearing_project)?;
    let bearing_root = bearing_project
        .and_then(|name| workspace.projects.get(name))
        .map(|p| &p.root);
    let targets = RelationalTargets::from_spec(workspace, spec);

    let mut selected: Vec<TaskId> = Vec::new();
    for candidate in &candidates {
        if !passes_non_relational(candidate, spec, bearing_root)? {
            continue;
        }
        let candidate_id = TaskId {
            project: candidate.project_name.clone(),
            task: candidate.task_name.clone(),
        };
        if !passes_relational(graph, &candidate_id, &targets) {
            continue;
        }
        selected.push(candidate_id);
    }
    Ok(selected)
}

/// Convenience wrapper for [`execute`] for specs that contain no
/// relational filters. Passes an empty graph; if any relational
/// filter is set on `spec`, that filter's target set will be
/// computed but the relation check will always fail (no edges to
/// traverse), producing zero matches. New code SHOULD use
/// [`execute`] directly.
///
/// # Errors
///
/// Same as [`execute`].
pub fn execute_non_relational(
    workspace: &Workspace,
    bearing_project: Option<&ProjectName>,
    spec: &QuerySpec,
) -> Result<Vec<TaskId>, QueryError> {
    let graph = TaskGraph::default();
    execute(workspace, &graph, bearing_project, spec)
}

#[cfg(test)]
mod tests {
    use std::collections::{BTreeMap, BTreeSet};
    use std::path::PathBuf;
    use std::str::FromStr;

    use haz_dag::edge::{Edge, EdgeKind};
    use haz_dag::graph::TaskGraph;
    use haz_domain::action::TaskAction;
    use haz_domain::env::EnvSettings;
    use haz_domain::mutex::{Mutex, MutexMode, MutexScope};
    use haz_domain::name::{MutexName, ProjectName, TagName, TaskName};
    use haz_domain::path::{
        CanonicalPath, HazPath, InputSpec, OutputSpec, PathPattern, ProjectRoot, WorkspaceRootPath,
    };
    use haz_domain::project::Project;
    use haz_domain::settings::WorkspaceSettings;
    use haz_domain::task::Task;
    use haz_domain::task_id::TaskId;
    use haz_domain::workspace::Workspace;
    use haz_query_lang::expr::{Expr, RawAtom};
    use haz_query_lang::span::Span;
    use nonempty::NonEmpty;

    use super::*;
    use crate::expr::relational::{RelationalAtom, parse_relational_atom};
    use crate::expr::shortcut::BooleanShortcut;

    fn argv(parts: &[&str]) -> NonEmpty<String> {
        NonEmpty::from_vec(parts.iter().map(|s| (*s).to_owned()).collect()).unwrap()
    }

    fn task(name: &str, inputs: &[&str], outputs: &[&str], with_mutex: bool) -> Task {
        Task {
            name: TaskName::from_str(name).unwrap(),
            action: TaskAction::Command(argv(&["true"])),
            inputs: inputs
                .iter()
                .map(|s| InputSpec::parse(s).unwrap())
                .collect(),
            outputs: outputs
                .iter()
                .map(|s| OutputSpec::parse(s).unwrap())
                .collect(),
            deps: vec![],
            weak_deps: vec![],
            mutex: if with_mutex {
                Some(Mutex {
                    scope: MutexScope::Workspace,
                    name: MutexName::from_str("db").unwrap(),
                    mode: MutexMode::Exclusive,
                })
            } else {
                None
            },
            env: EnvSettings::default(),
        }
    }

    fn project(name: &str, root: &str, tags: &[&str], tasks: Vec<Task>) -> Project {
        let mut task_map = BTreeMap::new();
        for t in tasks {
            task_map.insert(t.name.clone(), t);
        }
        Project {
            name: ProjectName::from_str(name).unwrap(),
            root: ProjectRoot::Nested(
                CanonicalPath::from_absolute(&HazPath::parse(root).unwrap()).unwrap(),
            ),
            tags: tags
                .iter()
                .map(|t| TagName::from_str(t).unwrap())
                .collect::<BTreeSet<_>>(),
            tasks: task_map,
        }
    }

    fn workspace(projects: Vec<Project>) -> Workspace {
        let mut map = BTreeMap::new();
        for project in projects {
            map.insert(project.name.clone(), project);
        }
        Workspace {
            root: WorkspaceRootPath::try_new(PathBuf::from("/abs/workspace")).unwrap(),
            projects: map,
            overlays: BTreeMap::new(),
            settings: WorkspaceSettings::default(),
        }
    }

    fn ids(names: &[(&str, &str)]) -> Vec<haz_domain::task_id::TaskId> {
        names
            .iter()
            .map(|(p, t)| haz_domain::task_id::TaskId {
                project: ProjectName::from_str(p).unwrap(),
                task: TaskName::from_str(t).unwrap(),
            })
            .collect()
    }

    fn tag_atom(s: &str) -> Expr<TagName> {
        Expr::Atom(TagName::from_str(s).unwrap())
    }

    fn project_atom(s: &str) -> Expr<ProjectName> {
        Expr::Atom(ProjectName::from_str(s).unwrap())
    }

    fn task_atom(s: &str) -> Expr<TaskName> {
        Expr::Atom(TaskName::from_str(s).unwrap())
    }

    fn pattern_atom(s: &str) -> Expr<PathPattern> {
        Expr::Atom(PathPattern::parse(s).unwrap())
    }

    fn three_project_workspace() -> Workspace {
        workspace(vec![
            project(
                "lib",
                "/lib",
                &["backend", "rust"],
                vec![
                    task("build", &["src/**/*.rs"], &["target/lib.so"], false),
                    task("test", &["src/**/*.rs"], &[], true),
                ],
            ),
            project(
                "web",
                "/web",
                &["frontend"],
                vec![task("bundle", &["src/index.ts"], &["dist/app.js"], false)],
            ),
            project(
                "tools",
                "/tools",
                &["backend"],
                vec![task("lint", &[], &[], false)],
            ),
        ])
    }

    // --- QRY-007: empty spec returns the candidate set --------

    #[test]
    fn qry_007_empty_spec_returns_full_workspace_candidate_set() {
        let ws = three_project_workspace();
        let selected = execute_non_relational(&ws, None, &QuerySpec::default()).unwrap();
        assert_eq!(
            selected,
            ids(&[
                ("lib", "build"),
                ("lib", "test"),
                ("tools", "lint"),
                ("web", "bundle"),
            ]),
        );
    }

    #[test]
    fn qry_007_empty_spec_with_bearing_project_restricts_to_that_project() {
        let ws = three_project_workspace();
        let bearing = ProjectName::from_str("lib").unwrap();
        let selected = execute_non_relational(&ws, Some(&bearing), &QuerySpec::default()).unwrap();
        assert_eq!(selected, ids(&[("lib", "build"), ("lib", "test")]));
    }

    // --- QRY-003 per-attribute (identifier) -------------------

    #[test]
    fn qry_003_tags_filter_selects_only_tagged_projects() {
        let ws = three_project_workspace();
        let spec = QuerySpec {
            tags: Some(tag_atom("backend")),
            ..QuerySpec::default()
        };
        let selected = execute_non_relational(&ws, None, &spec).unwrap();
        assert_eq!(
            selected,
            ids(&[("lib", "build"), ("lib", "test"), ("tools", "lint")]),
        );
    }

    #[test]
    fn qry_003_projects_filter_selects_only_matching_project() {
        let ws = three_project_workspace();
        let spec = QuerySpec {
            projects: Some(project_atom("web")),
            ..QuerySpec::default()
        };
        let selected = execute_non_relational(&ws, None, &spec).unwrap();
        assert_eq!(selected, ids(&[("web", "bundle")]));
    }

    #[test]
    fn qry_003_tasks_filter_selects_only_tasks_with_that_name() {
        let ws = three_project_workspace();
        let spec = QuerySpec {
            tasks: Some(task_atom("build")),
            ..QuerySpec::default()
        };
        let selected = execute_non_relational(&ws, None, &spec).unwrap();
        assert_eq!(selected, ids(&[("lib", "build")]));
    }

    #[test]
    fn qry_003_combined_per_attribute_filters_intersect() {
        let ws = three_project_workspace();
        let spec = QuerySpec {
            tags: Some(tag_atom("backend")),
            tasks: Some(task_atom("test")),
            ..QuerySpec::default()
        };
        let selected = execute_non_relational(&ws, None, &spec).unwrap();
        assert_eq!(selected, ids(&[("lib", "test")]));
    }

    // --- QRY-002 boolean operators in atoms ------------------

    #[test]
    fn qry_002_tag_expression_with_negation_evaluates() {
        let ws = three_project_workspace();
        // tags: backend & !frontend  -> backend-tagged but not frontend-tagged
        let spec = QuerySpec {
            tags: Some(Expr::And(
                Box::new(tag_atom("backend")),
                Box::new(Expr::Not(Box::new(tag_atom("frontend")))),
            )),
            ..QuerySpec::default()
        };
        let selected = execute_non_relational(&ws, None, &spec).unwrap();
        assert_eq!(
            selected,
            ids(&[("lib", "build"), ("lib", "test"), ("tools", "lint")]),
        );
    }

    // --- QRY-003 inputs / outputs with canonicalisation -------

    #[test]
    fn qry_003_inputs_workspace_absolute_atom_matches_canonicalised_task_pattern() {
        let ws = three_project_workspace();
        let spec = QuerySpec {
            inputs: Some(pattern_atom("/lib/src/main.rs")),
            ..QuerySpec::default()
        };
        let selected = execute_non_relational(&ws, None, &spec).unwrap();
        // lib's build + test both declare `src/**/*.rs`, which
        // canonicalises to `/lib/src/**/*.rs` and matches the
        // user's literal.
        assert_eq!(selected, ids(&[("lib", "build"), ("lib", "test")]));
    }

    #[test]
    fn qry_003_inputs_disjoint_workspace_root_excludes_other_projects() {
        let ws = three_project_workspace();
        let spec = QuerySpec {
            inputs: Some(pattern_atom("/web/src/index.ts")),
            ..QuerySpec::default()
        };
        let selected = execute_non_relational(&ws, None, &spec).unwrap();
        assert_eq!(selected, ids(&[("web", "bundle")]));
    }

    #[test]
    fn qry_003_outputs_glob_atom_matches_literal_task_output() {
        let ws = three_project_workspace();
        let spec = QuerySpec {
            outputs: Some(pattern_atom("/lib/target/*.so")),
            ..QuerySpec::default()
        };
        let selected = execute_non_relational(&ws, None, &spec).unwrap();
        assert_eq!(selected, ids(&[("lib", "build")]));
    }

    #[test]
    fn qry_003_inputs_project_relative_atom_with_bearing_project() {
        let ws = three_project_workspace();
        let bearing = ProjectName::from_str("lib").unwrap();
        let spec = QuerySpec {
            // User typed `src/main.rs` from inside `lib`.
            inputs: Some(pattern_atom("src/main.rs")),
            ..QuerySpec::default()
        };
        let selected = execute_non_relational(&ws, Some(&bearing), &spec).unwrap();
        assert_eq!(selected, ids(&[("lib", "build"), ("lib", "test")]));
    }

    #[test]
    fn qry_003_inputs_project_relative_atom_without_bearing_project_errors() {
        let ws = three_project_workspace();
        let spec = QuerySpec {
            inputs: Some(pattern_atom("src/main.rs")),
            ..QuerySpec::default()
        };
        let err = execute_non_relational(&ws, None, &spec).unwrap_err();
        match err {
            QueryError::CanonicalisePattern { canonical } => assert_eq!(canonical, "src/main.rs"),
            other => panic!("expected CanonicalisePattern, got {other:?}"),
        }
    }

    // --- QRY-005 shortcuts ------------------------------------

    #[test]
    fn qry_005_no_inputs_shortcut_selects_input_less_tasks() {
        let ws = three_project_workspace();
        let spec = QuerySpec {
            shortcuts: vec![BooleanShortcut::NoInputs],
            ..QuerySpec::default()
        };
        let selected = execute_non_relational(&ws, None, &spec).unwrap();
        assert_eq!(selected, ids(&[("tools", "lint")]));
    }

    #[test]
    fn qry_005_mutex_shortcut_selects_only_mutex_tasks() {
        let ws = three_project_workspace();
        let spec = QuerySpec {
            shortcuts: vec![BooleanShortcut::Mutex],
            ..QuerySpec::default()
        };
        let selected = execute_non_relational(&ws, None, &spec).unwrap();
        assert_eq!(selected, ids(&[("lib", "test")]));
    }

    #[test]
    fn qry_006_combining_multiple_filter_families_intersects() {
        let ws = three_project_workspace();
        let spec = QuerySpec {
            tags: Some(tag_atom("backend")),
            shortcuts: vec![BooleanShortcut::Mutex],
            ..QuerySpec::default()
        };
        let selected = execute_non_relational(&ws, None, &spec).unwrap();
        assert_eq!(selected, ids(&[("lib", "test")]));
    }

    // --- QRY-004 relational filters ---------------------------

    fn task_id(project: &str, task: &str) -> TaskId {
        TaskId {
            project: ProjectName::from_str(project).unwrap(),
            task: TaskName::from_str(task).unwrap(),
        }
    }

    fn relational_atom(text: &str) -> Expr<RelationalAtom> {
        Expr::Atom(
            parse_relational_atom(RawAtom {
                text: text.to_owned(),
                span: Span { start: 0, end: 0 },
            })
            .unwrap(),
        )
    }

    /// Graph aligned with [`three_project_workspace`]:
    /// `lib:build -> lib:test`, `lib:build -> web:bundle`,
    /// `lib:test -> tools:lint`.
    fn three_project_hard_edge_graph() -> TaskGraph {
        let pairs = [
            (task_id("lib", "build"), task_id("lib", "test")),
            (task_id("lib", "build"), task_id("web", "bundle")),
            (task_id("lib", "test"), task_id("tools", "lint")),
        ];
        let mut nodes: BTreeSet<TaskId> = BTreeSet::new();
        let mut edges: BTreeSet<Edge> = BTreeSet::new();
        for (from, to) in &pairs {
            nodes.insert(from.clone());
            nodes.insert(to.clone());
            edges.insert(Edge {
                from: from.clone(),
                to: to.clone(),
                kind: EdgeKind::Hard,
            });
        }
        TaskGraph { nodes, edges }
    }

    #[test]
    fn qry_004_child_of_selects_direct_successors_of_target_set() {
        let ws = three_project_workspace();
        let graph = three_project_hard_edge_graph();
        let spec = QuerySpec {
            child_of: Some(relational_atom("name:build")),
            ..QuerySpec::default()
        };
        let selected = execute(&ws, &graph, None, &spec).unwrap();
        assert_eq!(selected, ids(&[("lib", "test"), ("web", "bundle")]));
    }

    #[test]
    fn qry_004_parent_of_selects_direct_predecessors_of_target_set() {
        let ws = three_project_workspace();
        let graph = three_project_hard_edge_graph();
        let spec = QuerySpec {
            parent_of: Some(relational_atom("name:test")),
            ..QuerySpec::default()
        };
        let selected = execute(&ws, &graph, None, &spec).unwrap();
        assert_eq!(selected, ids(&[("lib", "build")]));
    }

    #[test]
    fn qry_004_depends_on_traverses_transitively() {
        let ws = three_project_workspace();
        let graph = three_project_hard_edge_graph();
        let spec = QuerySpec {
            depends_on: Some(relational_atom("name:build")),
            ..QuerySpec::default()
        };
        let selected = execute(&ws, &graph, None, &spec).unwrap();
        // lib:test depends on lib:build directly; tools:lint
        // transitively (lib:build -> lib:test -> tools:lint);
        // web:bundle directly. lib:build itself is in target_set
        // and excluded.
        assert_eq!(
            selected,
            ids(&[("lib", "test"), ("tools", "lint"), ("web", "bundle")]),
        );
    }

    #[test]
    fn qry_004_ancestor_of_traverses_transitively() {
        let ws = three_project_workspace();
        let graph = three_project_hard_edge_graph();
        let spec = QuerySpec {
            ancestor_of: Some(relational_atom("name:lint")),
            ..QuerySpec::default()
        };
        let selected = execute(&ws, &graph, None, &spec).unwrap();
        // tools:lint <- lib:test <- lib:build, so lib:build and
        // lib:test are ancestors; tools:lint itself is in
        // target_set and excluded.
        assert_eq!(selected, ids(&[("lib", "build"), ("lib", "test")]));
    }

    #[test]
    fn qry_004_depends_on_excludes_tasks_in_target_set() {
        let ws = three_project_workspace();
        let graph = three_project_hard_edge_graph();
        let spec = QuerySpec {
            depends_on: Some(relational_atom("project:lib")),
            ..QuerySpec::default()
        };
        let selected = execute(&ws, &graph, None, &spec).unwrap();
        // target_set = {lib:build, lib:test}. lib:test depends on
        // lib:build, but is itself in target_set and so is
        // excluded by the "minus EXPR-matching tasks themselves"
        // clause. tools:lint and web:bundle stand.
        assert_eq!(selected, ids(&[("tools", "lint"), ("web", "bundle")]));
    }

    #[test]
    fn qry_004_relational_filter_intersects_with_per_attribute() {
        let ws = three_project_workspace();
        let graph = three_project_hard_edge_graph();
        let spec = QuerySpec {
            child_of: Some(relational_atom("name:build")),
            tags: Some(tag_atom("frontend")),
            ..QuerySpec::default()
        };
        let selected = execute(&ws, &graph, None, &spec).unwrap();
        // child_of name:build => {lib:test, web:bundle}.
        // tag:frontend => {web:bundle}.
        // Intersection: {web:bundle}.
        assert_eq!(selected, ids(&[("web", "bundle")]));
    }

    #[test]
    fn qry_004_empty_target_set_yields_zero_matches() {
        let ws = three_project_workspace();
        let graph = three_project_hard_edge_graph();
        let spec = QuerySpec {
            child_of: Some(relational_atom("name:absent")),
            ..QuerySpec::default()
        };
        let selected = execute(&ws, &graph, None, &spec).unwrap();
        assert!(selected.is_empty());
    }

    #[test]
    fn qry_004_relational_with_boolean_composition_in_atom() {
        let ws = three_project_workspace();
        let graph = three_project_hard_edge_graph();
        // child_of (name:build | name:test) => direct successors
        // of {lib:build, lib:test} = {lib:test, web:bundle,
        // tools:lint}.
        let spec = QuerySpec {
            child_of: Some(Expr::Or(
                Box::new(relational_atom("name:build")),
                Box::new(relational_atom("name:test")),
            )),
            ..QuerySpec::default()
        };
        let selected = execute(&ws, &graph, None, &spec).unwrap();
        assert_eq!(
            selected,
            ids(&[("lib", "test"), ("tools", "lint"), ("web", "bundle")]),
        );
    }

    #[test]
    fn qry_004_combined_relational_filters_intersect() {
        let ws = three_project_workspace();
        let graph = three_project_hard_edge_graph();
        // depends_on name:build => {lib:test, web:bundle, tools:lint}
        // ancestor_of name:lint => {lib:build, lib:test}
        // Intersection: {lib:test}.
        let spec = QuerySpec {
            depends_on: Some(relational_atom("name:build")),
            ancestor_of: Some(relational_atom("name:lint")),
            ..QuerySpec::default()
        };
        let selected = execute(&ws, &graph, None, &spec).unwrap();
        assert_eq!(selected, ids(&[("lib", "test")]));
    }

    #[test]
    fn execute_with_no_relational_filter_matches_non_relational_entry() {
        let ws = three_project_workspace();
        let graph = three_project_hard_edge_graph();
        let spec = QuerySpec {
            tags: Some(tag_atom("backend")),
            ..QuerySpec::default()
        };
        let with_graph = execute(&ws, &graph, None, &spec).unwrap();
        let without_graph = execute_non_relational(&ws, None, &spec).unwrap();
        assert_eq!(with_graph, without_graph);
    }
}